Senior AI Platform Engineer building production LLM systems, retrieval pipelines, AI data platforms, and backend workflow products.
I build production AI systems that have to survive real usage, not just demos. My work sits at the intersection of backend engineering, retrieval, secure data access, workflow automation, observability, and platform reliability.
At ExponentHR, I lead enterprise data platform modernization across ETL reliability, deployment automation, and multi-tenant analytics operations supporting 400 enterprise clients. I reworked CDC ETL from about 30 minutes to under 8 minutes, reduced compute cost by about 67%, compressed deployment cycles from 3 months to 14 days, and automated database refresh workflows that previously required heavy manual effort.
Outside work, I ship applied AI products across FastAPI, TypeScript, PostgreSQL, Chrome extensions, RAG pipelines, multi-provider model routing, packaging, CI/CD, and live deployments.
- AI platform engineering for enterprise or developer-facing AI products
- Applied AI systems with retrieval, model routing, evaluation loops, and secure data access
- Backend and platform work where performance, observability, cost, and reliability matter
- End-to-end ownership from architecture and implementation through production operations
|
At ExponentHR, I led ETL modernization, deployment automation, and database operations across a multi-tenant HR/payroll data platform serving enterprise clients. 400 clients | 30m -> under 8m ETL runtime | 3 months -> 14 days deployment cycle | 67% ETL compute reduction |
AutoApply AI is the workflow platform that connects discovery, tailoring, application automation, and tracking across Chrome MV3, FastAPI, retrieval, and model routing. discover -> tailor -> apply -> track | 11 ATS adapters | 355+ backend tests |
|
tailor-resume is the extracted intelligence layer behind that workflow, delivered through CLI, Streamlit, MCP, and Python package surfaces with strong test coverage. CLI | Streamlit | MCP | PyPI | 190 tests |
JobScout powers the discovery side with scraping, preference matching, relevance scoring, alerts, and application memory across 130+ companies. 130+ companies | ranking engine | alerts | preference matching |
|
FinTune is a production-grade financial NLP system: QLoRA fine-tuning on Mistral-7B, 4-bit quantized inference, PII guardrails, real-time monitoring with drift detection, and autonomous self-recovery via circuit breaker pattern. QLoRA | PEFT | 4-bit NF4 | FastAPI | circuit breaker | 35+ tests |
Fraud Detection is a real-time ML pipeline for transaction fraud detection with streaming inference, feature engineering, and model serving at scale. real-time | streaming | feature store | model serving |
- Governed enterprise data and platform systems
- LLM fine-tuning, quantization, and model optimization for production inference
- LLM backends and retrieval pipelines
- AI workflow products that connect discovery, generation, and action
- Production engineering for systems that need measurable quality, clear guardrails, and reliable operations
- Missouri S&T: built Azure AI anomaly detection pipelines, improved alert quality, migrated workloads to AKS with HPA, and published NLP research.
- Zomato: built competitor analytics, search relevance, and internal search systems at production scale.
- M.S. Information Science & Technology, Missouri S&T, 4.0 GPA
- DP-700 Microsoft Certified Data Engineer
- Published researcher: Sentiment Analysis for Visitor Insights
I build production AI systems, retrieval pipelines, and governed data platforms that make teams faster without making systems harder to trust.